Struct Collider

pub struct Collider<'a> {
    source_exprs: &'a [PartitionExpr],
    atomic_exprs: Vec<AtomicExpr>,
    normalized_values: HashMap<String, Vec<(Value, OrderedF64)>>,
}

Collider is used to collide a list of PartitionExpr into a list of AtomicExpr

It also normalizes the values of the columns in the expressions.

Fields

source_exprs: &'a [PartitionExpr]

atomic_exprs: Vec<AtomicExpr>

normalized_values: HashMap<String, Vec<(Value, OrderedF64)>>

A map of column name to a list of (value, normalized value) pairs.

The normalized value is used for comparison. The normalization process keeps the order of the values.

Implementations

impl<'a> Collider<'a>

pub fn new(source_exprs: &'a [PartitionExpr]) -> Result<Self>

fn collect_column_values_from_expr( expr: &PartitionExpr, values: &mut HashMap<String, Vec<Value>>, ) -> Result<()>

Helper to collect values with their associated columns from an expression

fn collide_expr( expr: &PartitionExpr, index: usize, normalized_values: &HashMap<String, HashMap<Value, OrderedF64>>, result: &mut Vec<AtomicExpr>, ) -> Result<()>

Collide a PartitionExpr into multiple AtomicExprs.

Split the PartitionExpr on every RestrictedOp::Or (disjunction), each branch is an AtomicExpr. Since PartitionExpr doesn’t allow parentheses, Expression like (a = 1 OR b = 2) AND c = 3 won’t occur. We can safely split on every RestrictedOp::Or.

fn collect_nucleons_from_expr( expr: &PartitionExpr, nucleons: &mut Vec<NucleonExpr>, normalized_values: &HashMap<String, HashMap<Value, OrderedF64>>, ) -> Result<()>

Collect nucleons from an expression (handles AND operations recursively)

fn collect_nucleons_from_operand( operand: &Operand, nucleons: &mut Vec<NucleonExpr>, normalized_values: &HashMap<String, HashMap<Value, OrderedF64>>, ) -> Result<()>

Collect nucleons from an operand

fn try_create_nucleon( lhs: &Operand, op: &RestrictedOp, rhs: &Operand, normalized_values: &HashMap<String, HashMap<Value, OrderedF64>>, ) -> Result<NucleonExpr>

Try to create a nucleon from operands